wisegroup

Now I am become death, destroyer of groups….

Never be burned by forgetting to ungroup() again.

This packages contains alternative versions of common group-aware tidyverse functions that either:

• automatically ungroup the returned data OR
• allow the developer to make explicit that groups are being carried forward, signalling the potential for group related errors with a ‘…’ variant of the orignal function, e.g. mutate..., summarise... etc.

Installation

remotes::install_github("milesmcbain/wisegroup")

Usage

wisegroup is designed to mask tidyverse functions in your global environment. It is recommended that you use this package in conjunction with conflicted. This means that the order of the library calls will not matter, and that the masking is made explicit like so:

library(conflicted)
library(wisegroup)
library(tidyverse)
#> ── Attaching packages ────────────────────────────────── tidyverse 1.2.1 ──
#> ✔ ggplot2 3.2.0     ✔ purrr   0.3.3
#> ✔ tibble  2.1.3     ✔ dplyr   0.8.3
#> ✔ tidyr   0.8.3     ✔ stringr 1.4.0
#> ✔ readr   1.3.1     ✔ forcats 0.4.0
#> ── Conflicts ───────────────────────────────────── tidyverse_conflicts() ──
#> ✖ dplyr::add_count()     masks wisegroup::add_count()
#> ✖ dplyr::add_tally()     masks wisegroup::add_tally()
#> ✖ dplyr::count()         masks wisegroup::count()
#> ✖ tidyr::fill()          masks wisegroup::fill()
#> ✖ dplyr::filter()        masks stats::filter()
#> ✖ dplyr::lag()           masks stats::lag()
#> ✖ dplyr::mutate()        masks wisegroup::mutate()
#> ✖ dplyr::mutate_all()    masks wisegroup::mutate_all()
#> ✖ dplyr::mutate_at()     masks wisegroup::mutate_at()
#> ✖ dplyr::mutate_if()     masks wisegroup::mutate_if()
#> ✖ dplyr::sample_frac()   masks wisegroup::sample_frac()
#> ✖ dplyr::sample_n()      masks wisegroup::sample_n()
#> ✖ dplyr::slice()         masks wisegroup::slice()
#> ✖ dplyr::summarise()     masks wisegroup::summarise()
#> ✖ dplyr::summarise_all() masks wisegroup::summarise_all()
#> ✖ dplyr::summarise_at()  masks wisegroup::summarise_at()
#> ✖ dplyr::summarise_if()  masks wisegroup::summarise_if()
#> ✖ dplyr::summarize()     masks wisegroup::summarize()
#> ✖ dplyr::summarize_all() masks wisegroup::summarize_all()
#> ✖ dplyr::summarize_if()  masks wisegroup::summarize_if()
#> ✖ dplyr::tally()         masks wisegroup::tally()
#> ✖ dplyr::transmute()     masks wisegroup::transmute()
#> ✖ dplyr::transmute_all() masks wisegroup::transmute_all()
#> ✖ dplyr::transmute_at()  masks wisegroup::transmute_at()
#> ✖ dplyr::transmute_if()  masks wisegroup::transmute_if()
library(nycflights13)

conflict_prefer("summarise", "wisegroup")
#> [conflicted] Will prefer wisegroup::summarise over any other package
conflict_prefer("mutate", "wisegroup")
#> [conflicted] Will prefer wisegroup::mutate over any other package

Just be lazy. conflicted will let you know when you need to disambiguate.

So now you have automatically dropped groups after summarising over multiple grouping columns:

grouped_no_more <-
  flights %>%
  group_by(year, month, day) %>%
  summarise(max_delay = max(dep_delay, na.rm = TRUE))

grouped_no_more
#> # A tibble: 365 x 4
#>     year month   day max_delay
#>    <int> <int> <int>     <dbl>
#>  1  2013     1     1       853
#>  2  2013     1     2       379
#>  3  2013     1     3       291
#>  4  2013     1     4       288
#>  5  2013     1     5       327
#>  6  2013     1     6       202
#>  7  2013     1     7       366
#>  8  2013     1     8       188
#>  9  2013     1     9      1301
#> 10  2013     1    10      1126
#> # … with 355 more rows

You indicate the grouping is sticking around with the ... suffix

still_grouped <-
  flights %>%
  group_by(year, month, day) %>%
  summarise...(max_delay = max(dep_delay, na.rm = TRUE))

still_grouped
#> # A tibble: 365 x 4
#> # Groups:   year, month [12]
#>     year month   day max_delay
#>    <int> <int> <int>     <dbl>
#>  1  2013     1     1       853
#>  2  2013     1     2       379
#>  3  2013     1     3       291
#>  4  2013     1     4       288
#>  5  2013     1     5       327
#>  6  2013     1     6       202
#>  7  2013     1     7       366
#>  8  2013     1     8       188
#>  9  2013     1     9      1301
#> 10  2013     1    10      1126
#> # … with 355 more rows

and with mutate:

summarised_over_mutates_groups <-
  flights %>%
  group_by(year, month, day) %>%
  mutate...(day_dep_variance = var(dep_delay, na.rm = TRUE)) %>%
  summarise(max_var = max(day_dep_variance))

summarised_over_mutates_groups
#> # A tibble: 365 x 4
#>     year month   day max_var
#>    <int> <int> <int>   <dbl>
#>  1  2013     1     1   2049.
#>  2  2013     1     2   1384.
#>  3  2013     1     3    990.
#>  4  2013     1     4    769.
#>  5  2013     1     5    662.
#>  6  2013     1     6    537.
#>  7  2013     1     7    733.
#>  8  2013     1     8    374.
#>  9  2013     1     9   2197.
#> 10  2013     1    10   1924.
#> # … with 355 more rows

Fun Fact

This package generates its own source and documentation, so that adding new wrappers is low effort.

To generate new wrappers, an entry is added to “./group_aware_functions.R” and “./make.R” is run.

Contributing

If there’s a function you think should be included make a PR for “./group_aware_functions.R”!